Ellis, thank you for the long and educated conversation. Despite our intellectual disagreement, I genuinely appreciate the rigor you’ve brought to this debate. If you’re at Bespoked in London in April, the first craft beer is on me.
We’re not as far apart as it might seem. I agree that scale, systematization, and process optimization change the nature of work. I agree that fixtures and procedures relocate judgement. I agree that distinguishing craft from manufacturing matters. Where we diverge is on the fundamental definitions themselves, and specifically, on treating Pye and Sennett as making the same argument. They’re not. In fact, they’re opposed on the essentials.
Sennett’s framework, which is central to my PhD research, comes from Hannah Arendt’s distinction between homo faber and animal laborans. Homo faber, the maker, understands and controls the entire domain of work from conception through execution to evaluation. The maker exercises judgement over the complete process, understands materials and constraints, and maintains responsibility for outcomes. Animal laborans, the laborer, is reduced to fragmented tasks without understanding the whole. They execute instructions, repeat procedures, but don’t control design decisions or evaluate long-term consequences.
This is not the same as Pye’s workmanship of risk versus workmanship of certainty. Pye’s framework is narrower, focused on where control resides during physical execution. Hand control equals risk, machine preset equals certainty. But this misses what Sennett and Arendt argue is fundamental, which is scope of judgement and domain control, not execution method.
Here’s where Pye’s framework breaks down. Pye labels traditional hand work as workmanship of risk, implying that risk-taking and uncertainty are central to craft. But traditional craft apprenticeship is built on the opposite: repetition, imitation, adherence to established methods, and incremental refinement within known constraints. That’s not risk. That’s incremental innovation, safe, conservative, validated by tradition. A framebuilder learning traditional brazing isn’t venturing into unknown territory. They’re replicating techniques passed down over generations, following established procedures, working within proven parameters. The execution varies slightly because hands aren’t perfectly consistent, but the approach is deeply risk-averse.
Meanwhile, Pye concepts dismisse digital fabrication as workmanship of certainty, as if precision execution means no judgement, no risk, no craft knowledge. But developing new fabrication systems and processes involves genuine uncertainty, experimental iteration, and real potential for failure. There’s no established tradition guaranteeing success. That’s disruptive innovation, actually risky, not just variable. By Pye’s own logic, if risk defines craft value, then work with new tools should be more craft-worthy than traditional apprenticeship, because it actually involves risk rather than just execution variability within known parameters. But Pye doesn’t go there, because his framework isn’t really about risk. It’s about preserving the legitimacy of traditional methods by mislabeling execution variability as risk while dismissing new approaches as merely mechanical.
Pye’s framing of hand work as inherently risky actually contradicts the learning-by-repetition model he’s supposedly defending. If craft is built on repetition and incremental improvement, then skilled craftspeople aren’t working under constant risk—they’re achieving consistent, predictable outcomes through practiced technique. That’s not risk, that’s developed competence. Meanwhile, his workmanship of certainty assumes tools, whether analog jigs or digital systems, produce perfectly predictable outcomes, which anyone who’s worked with precision tooling knows is false. Tools require calibration, materials vary, environmental conditions affect results, and developing reliable processes with any tool involves extensive trial and error. Pye’s framework creates a false binary where hand equals human equals fallible equals risky, and tool equals machine equals perfect equals certain. Neither half of that equation holds up to scrutiny.
His framework completely breaks down when confronted with mass customization. Pye assumed that customization requires hand control, workmanship of risk, while standardization enables machine control, workmanship of certainty. But digital fabrication tools enable mass customization, producing unique, individually specified objects at scale with precision and reliability. A framebuilder today can produce 200 bespoke frames per year using digital tools, achieving exactly what master framebuilders did historically through hand work. By Pye’s execution method, this is workmanship of certainty. By his customization criterion, it’s workmanship of risk. By actual practice, it’s the same craft activity framebuilders have always done. Pye’s framework can’t accommodate this because it conflates the technological constraints of 1968 with fundamental properties of craft. The mass customization point is devastating because it shows Pye’s framework isn’t just philosophically weak, it’s empirically obsolete. Master framebuilders were always doing mass customization. If digital tools can achieve the same outcome, then either the tools don’t matter, which kills Pye’s framework, or historical framebuilding wasn’t craft, which is absurd.
Sennett’s framework doesn’t have this problem. A framebuilder using digital tools can absolutely be homo faber if they understand materials, geometry, and structural principles, control the design systems that define the frame, make design decisions informed by performance requirements and rider feedback, evaluate outcomes and iterate based on real-world consequences, and maintain responsibility for the complete process from concept to delivery. They’re exercising craft judgement over the entire domain while relocating some execution to tools that provide precision and consistency. This is the bespoke model, maker controls the complete domain: design, fabrication, evaluation, customer relationship, iteration based on outcomes. That’s homo faber, regardless of tools used. Conversely, an apprentice in a traditional workshop who follows instructions without understanding principles, replicates techniques without questioning why, and never sees long-term outcomes is closer to animal laborans despite working by hand. They control execution but don’t understand or control the broader process. This is the industrial model, worker controls only a fragment with no input into design, no feedback from outcomes, no understanding of the whole. That’s animal laborans, whether done by hand or machine.
There’s an underlying assumption in craft theory, drawing from Polanyi’s work on tacit knowledge, that craft judgement is somehow irreducibly tacit, that it can’t be formalized, only transmitted through long apprenticeship and embodied practice. But Polanyi’s famous example, we know how to ride a bicycle but can’t explain the physics, actually undermines this claim. We teach millions of people to ride bicycles efficiently, often in hours or days, without needing to articulate the physics. The tacit knowledge transmits easily through structured practice and feedback. The question isn’t whether craft knowledge contains tacit elements. It does. The question is whether we want to transmit it, and to whom.
When craft knowledge remains purely tacit, never documented or formalized, it serves gatekeeping functions. Formalized knowledge doesn’t eliminate intuition, it makes foundations explicit so learners can develop intuition faster and more systematically. A framebuilder who understands tube geometry principles develops informed intuitions about design. A framebuilder who only replicates what they were shown develops habit, not understanding. Knowledge that cannot be examined, validated, and transmitted isn’t just tacit, it’s gatekept. When today’s economic conditions no longer support long, low-paid apprenticeships, insisting on purely tacit transmission doesn’t preserve craft. It makes craft accessible only to those with the economic privilege to spend years learning without income.
This brings me to my central thesis, which is what I call hybrid craftsmanship. This concept is fully aligned with Sennett’s framework of homo faber. Hybrid craftsmanship means maintaining complete domain control over the framebuilding process while integrating digital tools and formalized knowledge systems that extend rather than replace craft judgement. It’s the convergence of traditional understanding of materials, geometry, and structural principles with contemporary tools that enable precision, documentation, and systematic knowledge transmission. The craftsperson as homo faber doesn’t lose control or understanding by using digital design any more than they lose it by using a micrometer instead of eyeballing measurements. The tools change, the scope of judgement and responsibility remains. Hybrid craftsmanship recognizes that craft knowledge has both tacit and explicit dimensions, and that making the explicit dimensions accessible through formalization doesn’t diminish the tacit dimensions, it accelerates their development. It recognizes that volume matters for judgement formation, which is why B2B production structures remain viable for craft practice, not just for manufacturing. And it recognizes that economic conditions have changed in ways that make purely tacit, purely hand-based transmission models inaccessible to most people who want to learn framebuilding today.
If craft judgement requires exposure to consequences through volume, as you’ve argued and I agree, then B2B production provides the structure for that volume. And if craft knowledge is to be transmitted and preserved rather than dying with individual practitioners, then formalization makes that knowledge accessible and examinable rather than purely tacit and gatekept. Neither of these moves reduces homo faber to animal laborans. They maintain domain control and systematic learning while adapting to economic conditions that no longer support long, low-paid apprenticeships.
I’m doing this work with Sennett’s framework, which I believe is more robust than Pye’s because it doesn’t conflate execution method with scope of judgement. Pye’s framework serves to protect traditional methods by defining craft in terms that exclude innovation. Sennett’s framework asks whether the maker maintains understanding and control over the complete domain of work, which is the question that actually matters for craft preservation. If Pye’s theory fails, it’s because it mislabels conservative incrementalism as risk while dismissing genuine innovation as certainty, and because it privileges execution variability over domain control. Sennett provides the framework that replaces it.
Thanks again for pushing this conversation to definitional bedrock. That’s where the real work happens.
